Evolution and Arithmetic

Chapter 15: The Brain

← The Hardware Is the SoftwareRules and Specifics →

How Is a Brain Made?

Everything so far has been about what a brain is and does once it exists. Now comes the question this book actually needs answered: how does a brain come to exist at all, built by nothing but DNA?

The obvious first guess is that DNA holds the finished wiring diagram, connection by connection, the way an architect's drawings show every wire in a building. That turns out to be impossible, and not just unlikely. The arithmetic is simple. Merely writing down which neuron each of the brain's hundred trillion connections leads to, ignoring everything else a real wiring diagram would need, takes something like four billion phone books. The entire human genome, every letter of it, would fill about two and a half thousand. The wiring list is roughly six hundred thousand times bigger than all the instructions the body has to work with, and the genome has a great many other jobs besides. (The arithmetic: Wiring the Brain.) No improvement in the data will close a gap like that. Neuroscientists have a name for it, the "genomic bottleneck," argued most directly by Anthony Zador of Cold Spring Harbor Laboratory.

The conclusion Zador and others draw is the one this book reached from a different direction in Chapter 13: the genome can't be a blueprint, because a blueprint that detailed wouldn't fit. It has to be a compressed set of rules, general instructions for how to build and connect, not an itemized map of the result. A rule like "grow toward this chemical signal" or "connect to cells carrying a matching marker" can, in principle, produce an enormously complicated structure from a small amount of instruction, the way "take a step, turn a little, and make each step slightly longer than the last" traces out a spiral without the rule needing to name every point along it.

The leading idea of what such a rule looks like physically is a molecular address tag. Each neuron carries a particular combination of markers on its surface, and connections form where the right tags meet matching receptors. The idea goes back to the neuroscientist Roger Sperry in 1963, and it has real experimental support. But it's worth being honest about how far that support reaches. Even in C. elegans, with its wiring fully mapped, researchers say they haven't cracked the code that turns the worm's tags into its actual, specific wiring. There's no simpler case to fall back on. Three hundred and two neurons is as far down the ladder as this problem goes, and it's still unsolved there.

Last revised 2026-11-18

← The Hardware Is the SoftwareRules and Specifics →